Effects of measurement error and sampling resolution on estimates of atrial tissue recovery parameters.

Abstract

We studied the effect of sampling resolution and measurement error on estimates of tissue recovery parameters using experimental and simulated data. Action potential duration (APD) was estimated from monophasic action potentials recorded at 250 sites (delta x = 3.5 mm) on the endocardium of the canine right atrium (n = 8) during control and acetylcholine perfusion. APD distributions were also simulated using a random number generator, then scaled and filtered to physiological values. The following parameters were estimated at increasing APD sampling interval and measurement error: mean APD, standard deviation of APD, mean APD gradient, standard deviation of APD gradient, APD wavelength, and APD correlation length. We found that large errors can result from APDs collected at inadequate sampling intervals and adequate sampling intervals may be 3-6 times less than the Nyquist interval. Large parameter errors also resulted from data with relatively low levels of measurement error. The effect of measurement error was dependent upon the standard deviation of APD, sampling resolution, and APD wavelength. Inadequate sampling resolution was the largest source of error in experimental parameter estimates. Estimates of mean and standard deviation of APD gradient decreased with spacing as estimates of correlation length and wavelength increased. Careful selection of spacing interval, taking into account the spatial complexity of recovery, as well as considerably low measurement errors will produce accurate estimates of gradients, correlation length, and wavelength.